Overview of Meteorological Surface Variables and Boundary-layer Structures in the Seoul Metropolitan Area during the MAPS-Seoul Campaign

The meteorological surface variables and boundary-layer structures in the Seoul Metropolitan Area (SMA) were examined during the MAPS-Seoul (Megacity Air Pollution Studies in Seoul) intensive field experiment (18 May–13 June 2015). Data were obtained from a surface energy balance system and a ceilometer installed at a UMS-Seoul (urban meteorological observation system in the SMA) station located in the eastern part of the city of Seoul. A series of migratory anticyclone passages under a strong El Niño event meant that the experimental period was much warmer and drier than the climatological ones. The mean air temperature during this period was 2.6°C higher than the recent 30-year (1981–2010) climatological value, while only one-eighth of the climatological precipitation was recorded. Twelve mist event days were recorded, as were four of haze and six of rainfall events. The SMA was found to be affected by the sea-land breeze: Westerly winds (sea-breezes) were dominant from the afternoon till the early evening, while easterlies (land-breezes) were dominant before the morning. Finally, the vertical profile of the attenuated backscatter obtained by the ceilometer indicated that the maximum daily mixing-layer heights (MLHs) on days with mist/haze and much cloud cover were lower than on days with no-mist/haze and less cloud cover, respectively, mainly due to a decrease in the downward solar radiation. However, the effect of mist/haze on the daytime MLH evolution was larger than that of cloud cover. The MLH also exhibited an altitude similar to that of the steepest vertical gradient of sulfate and organic aerosol concentration, obtained from the airborne measurement on 13 June 2015.


INTRODUCTION
East Asia is a major source of natural and anthropogenic aerosols over the Northern Hemisphere (Park et al., 2010a(Park et al., , 2013a)).Asian dust is an example of a natural mineral aerosol, which occurs in Northern China and Mongolia, more frequently during the spring season (Husar et al., 2001;In and Park, 2003;Park et al., 2010b).Anthropogenic aerosols originate from human activities, and are formed by gas-to-particle conversion processes in the atmosphere.They are abundant due to the high pollutant emissions associated with rapid economic expansion in East Asia (Streets et al., 2003;Zhang et al., 2009).
Korea is located in the downwind region of China, one of the world's largest emitters of air pollutants.Although there were many previous studies on the air quality model and observation in this region (Koo et al., 2008;Park et al., 2010b;Han et al., 2011;Jeong et al., 2013;Park et al., 2015;Chuang et al., 2017;Kim et al., 2017;Seo et al., 2017), the quantitative determination of the main contributors to atmospheric pollution in Korea, not only during high air pollution events but also during longer-term periods, has still been controversial.
The Megacity Air Pollution Studies in Seoul (MAPS-Seoul) project was launched to improve the prediction performance of air quality models in Eastern Asia.The intention of the project was to reduce the model's bias, and thus to assist the formulation of long-term air quality policy in East Asia as a whole, and in Korea in particular (NIER, 2015a;Kim et al., 2017;Park et al., 2018).For this purpose, the MAPS-Seoul intensive field experiment campaign was designed and conducted during the period from 18 May to 13 June 2015.It integrated all kinds of air quality and aerosol observation from multi-platforms as well as the results of meteorological and air quality models.The MAPS-Seoul campaign subsequently continued as the Korea-United States Air Quality Study (KORUS-AQ) campaign in 2016, jointly organized by the National Institute of Environmental Research (NIER) of Korea and the National Aeronautics and Space Administration (NASA) of United States (https://espo.nasa.gov/korusaq/content/KORUS-AQ).
The MAPS-Seoul campaign involved the collection of not only ground-based, air-borne, and satellite-based aerosol and air pollutant measurements.There were 7 airborne aerosol concentration measurement in the King Air of Hanseo University (NIER, 2015b).Although surface and vertical profiles of meteorological variables could play very important roles in understanding air quality processes and simulating air pollutant concentrations more accurately, only the limited meteorological variables were observed.Fortunately, a high-resolution urban meteorological observation system network (UMS-Seoul) has been installed in the Seoul Metropolitan Area (SMA) recently.Its purpose was to prevent possible damage from unexpected extreme disasters, and to give a timely guidance to citizen by delivering high-quality meteorological information customized for users' demands (Park et al., 2017).
This study aims to provide an overview of the climatological features observed during the MAPS-Seoul campaign in SMA.Then meteorological surface variables and boundary-layer structures are to be discussed with the use of data observed by a surface energy balance system and a ceilometer in the SMA.Finally, the effects of mist/haze and cloud cover on the evolution of mixing-layer height and nighttime mixing-layer height are to be analyzed.

Stations
The Jungnang Station was selected from among the major stations of the UMS-Seoul.The Station had many meteorological observation platforms such as surface meteorological variables, a surface energy balance (turbulence or flux), vertical profiles of temperature, moisture, wind, and aerosols (Park et al., 2017).The station is located in the eastern part of Seoul City, and is about 40 km to the east of the Yellow Sea (Fig. 1(a)).It was installed on the rooftop of a building located in an urban residential area.The mean building height in the area was 11.0 m, the plane area density was 0.52, and the frontal area density was 0.44 (Fig. 1(b); Kwon et al., 2014).The data obtained at the station was found to well represent the meteorological feature of Seoul.The climatic data and precipitation data obtained at the Seoul Station were used.The Seoul Station is an Automatic Synoptic Observation Station (ASOS) operated by the Korea Meteorological Administration.It is located at the center of Seoul City and 10 km to the west of the Jungnang Station (Fig. 1(a)).
Hourly SO 2 , CO, O 3 , NO 2 , PM 10 , and PM 2.5 concentrations were measured at the air quality monitoring station (station number 111151), operated by the Korean Environment Corporation, funded by the Ministry of Environment.The station is located in the same Jungnang administrative district of Seoul City, and 1.9 km to the south of the Jungnang Station (Fig. 1(a)).

Instrumentation
A surface energy balance system and a ceilometer were installed at the Jungnang Station.A surface energy balance system deployed temperature and humidity sensors at three levels, wind speed and direction sensors at three levels, 3dimensional sonic anemometers at three levels, CO 2 /H 2 O infrared gas analyzers at three levels, an air pressure sensor, a precipitation gauge with heater, and a four-component net radiometer (downward/upward and shortwave/longwave radiometer) on the 18.5 m meteorological observation tower.Meteorological variables were sampled every second and saved every minute, while turbulence variables (measured by sonic anemometers and CO 2 /H 2 O infrared gas analyzers) were sampled and saved at every 10 Hz.Turbulent components for computing the momentum, sensible heat, latent heat, and CO 2 fluxes were extracted with the use of the 30-minute block average (Park et al., 2014).
A ceilometer (manufactured by Vaisala, model CL51) was installed at the same station.It provided the vertical distribution of two-way attenuated backscatter by means of a 910 nm wavelength laser as well as three levels of cloud base height.Attenuated backscatter had a vertical resolution of 10 m up to 15 km and a temporal resolution of 1 minute.
The airborne measurements observed the organic aerosol, sulfate, nitrate, and ammonium concentrations during the 7 research flight (RF) periods.Each RF was conducted on a different route during the 2 to 3.5 hours in daytime: The RF01 (27 May) had a route in the south of Seoul; the RF02 (2 June) had a route in the Yellow Sea with different altitude; the RF03 (6 June) and RF06 (13 June) had routes in the Yellow Sea with spiral flight; the RF04 (7 June) and RF07 (13 June) flew across the Seoul and Wonju; the RF05 (7 June) flew across the Gwangju and Busan (NIER, 2015b).The flight times were indicated in Table 1 and Figs. 2,4,5,6,and 8.

Air Quality Variables
The temporal variation features of major air pollutants such as SO 2 , CO, O 3 , NO 2 , PM 10 , and PM 2.5 were briefly overviewed.Fig. 2 shows the time series of air pollutants concentration and visibility during the MAPS-Seoul campaign.The PM 10 concentration exhibited a similar variation pattern to PM 2.5 concentration, but the former with a mean of 39.5 µg m -3 was higher than the latter with a mean of 21.9 µg m -3 (Fig. 2(a)).Daily mean PM 10 concentrations on 19 and 22 May exceeded the daily environmental standard of 50 µg m -3 .High PM 10 concentration > 100 µg m -3 was recorded during the RF07 period on 13 June 2015.The PM 10 (PM 2.5 ) concentration was inclined to slightly increase (decrease) with time of day, thus the ratio of PM 2.5 to PM 10 with a mean of 55.5% was inclined to decrease with time of day (Fig. 2(d)).The SO 2 and CO concentrations were much lower than the environmental standard of 0.05 ppm (24-hour average) and  (23,27,28 May,6,7,10,11 June) among 26 days (Fig. 2(c)) and during the RF01, RF03-05 periods.On those days, daily maximum downward solar radiations were all larger than 840 W m -2 .The increase in O 3 concentration during daytime was mainly due to the increase in solar radiation and NO 2 concentration.On the while, the decrease in O 3 concentration during the early night was due to the reaction with NO and NO 2 (Lal et al., 2000;Ryu et al., 2013).The NO x was emitted from combustion engines by the road traffic (Niemeier et al., 2006;Park et al., 2014).As a result, the NO 2 concentration exhibited two maxima on the late morning hour and on midnight, and two minima on the intermediate hours (Fig. 2(c)).The lower visibilities than 1000 m were recorded on 11 days in the Seoul Station (Table 1).The 1000 m is the threshold distance for determination of the mist or haze event (Fig. 2(e)).The lower visibility events were occurred during the high PM and the high ratio of PM 2.5 to PM 10 periods (Figs.2(a) and 2(d)) (Park et al., 2013a).

Climatological Features and Meteorological Surface Variables
Korea's climate is classified as temperate with four distinct seasons.It has hot and humid summers due to the North Pacific high pressure system, and cold and dry winters due to the Siberian high pressure system (Jung et al., 2002).The Korean Peninsula was affected by a very strong El Niño starting in May 2015 and ending in February 2016 (WMO, 2016), and governed by an abnormally expanded North Pacific High, which stretched to 40°N latitude in the Western Pacific region (Fig. 3).And a summer monsoonal front was located in 20-30°N latitude throughout the period.These implied that the Korean Peninsula could be governed by warmer and drier weather than the climatology during the period.
Weather charts show that 11 migratory anticyclones and 5 cyclones passed the Korean Peninsula during the period.The pressure difference (anticyclone intensity) between the center and the point with the lowest pressure was relatively small, so it was not easy to clearly discriminate each anticyclone.Moreover, two or three anticyclones aligned from the middle of China, via the Korean Peninsula, to northern Japan along the northern border of the North Pacific High were stagnant from 2100 LST 22 to 0900 LST 25 May 2015 (Fig. 3).
Time series of air temperature and precipitation data observed during the MAPS-Seoul campaign were compared with the climatological records for the 30 years from 1981 to 2010.These showed that the campaign period was much warmer and drier compared to the climatology.The mean air temperature for this period was 22.7°C, 2.6°C higher than the climatological mean of 20.1°C (Fig. 4(a)).On 21 days (80%) among total 26 days, the daily maximum air temperatures exceeded the climatological mean.In particular, the daily maximum air temperature exceeded the 30-yearaveraged ones by more than 4°C for 6 consecutive days from 24 to 29 May.At the Seoul Station, very little precipitation (10.9 mm) was recorded compared with the climatological one (84.4mm) (Fig. 4(b)).Table 1 shows the daily cloud cover and weather phenomena which occurred at the Seoul Station during the MAPS-Seoul campaign.In total, 6 rainfall days, 12 mist days, and 4 haze days were recorded.The mist and haze events often occurred on the same days (25, 31 May, 10, 12 June).This was because a mist event was defined as occurring when relative humidity exceeded 80%, while a haze event was defined as occurring when relative humidity was less than 70% under the same visibility range from 1 to 10 km (Park et al., 2013a).Also, the rainfall and mist/haze events did sometimes occur on the same days (19,30 May,11,12,13 June), because a rainfall event was frequently accompanied by visibility reduction.For further study, the day with cloud cover less than or equal to 30% was defined Figs.5(a) and 5(b) show the time series for wind speed and direction, respectively.These measurements were made at 18 m high of the meteorological observation tower.This time series indicated that the winds exhibited a roughly diurnal variation with larger speeds and westerlies in daytime and smaller speeds and easterlies in nighttime.It was well-known that the local winds are dominant under the clear sky and weak synoptic system.In order to more clearly discriminate the features of local circulation, the winds were averaged on only 12 clear days (Table 1).Zonal and meridional mean wind speeds during the clear days were found to be 1.16 m s -1 and 0.05 m s -1 , respectively.It meant that the prevailing winds were westerly at the station.Clear-day mean diurnal variation showed that the westerly winds were dominant all day except for between 0400 and 0900 LST.The mean wind speed exhibited a maximum when westerlies were recorded in the late afternoon, and low values when easterly winds occurred before noon (Fig. 5(c)).In order to find out the local wind component from the observed one, wind anomalies were obtained by subtracting the clear day mean value from the observed winds.The resultant wind anomalies distinctively showed the existence of typical local wind patterns: westerly winds (sea breeze) were dominant from 1400 to 2000 LST and easterly winds (land breeze or urban-rural breeze) were dominant from 0000 to 1130 LST.The wind speed had two maxima near 1800 LST for westerly and 0600 LST for easterly, and two minima near 1300 LST and 2300 LST.Westerly anomaly winds meant that the station could be affected by sea-breeze or other local circulation, while easterly anomaly winds meant that the station could be affected by land-breeze, urban-rural breeze, or other local breeze.But the observed wind was the sum of synoptic wind and the local wind, and the local wind included the sea-land breeze, urban-rural breeze, and mountain-valley breeze, it is difficult to extract the urbanrural breeze contribution from observed wind.
To investigate the occurrence frequency of wind with respect to the time of a day, wind roses for the afternoon (1200-1800 LST) and the nighttime (0000-0600 LST) were drawn in Figs.5(e) and 5(f), respectively.When wind speed was less than 0.5 m s -1 , the wind was classified as "calm".The occurrence frequency for calm and the lowestclass wind speeds, between 0.5 m s -1 and 2 m s -1 , was 7.4% and 76.2% in the nighttime, respectively.These were much higher than those in the afternoon when calm and the lowest-class wind speeds were recorded with a frequency of 1.3% and 19.2%, respectively.The wind roses showed that 65.4% of winds were westerly (from south-west to west-north-west) in the afternoon, while 33.7% of winds were north-easterly (from north-north-east to east-northeast) during the nighttime (Figs.5(e) and 5(f)).Wind speeds which exceeded 3.4 m s -1 occurred as frequently as 44.9% during the afternoon, but only with a frequency of 4.8% in the nighttime (Figs.5(e) and 5(f)).This might support the previous model studies which determined that sea-breezes could reach inland regions approximately 50 km from the shoreline in the SMA (Ryu and Baik, 2013;Ganbat et al., 2015).
The solar and terrestrial radiations affect the concentration of air pollutants through the photochemical process as well as the vertical mixing in the boundary-layer.Station during the MAPS-Seoul campaign.Shortwave (solar) and longwave (terrestrial) radiation can indirectly give information on the horizontal and vertical distribution of clouds and aerosols.Temporal variations in radiation were found to be coincident with the cloud cover listed in Table 1.Note that the downward solar radiation decreases steeply on the rainy days (Fig. 6(a)).Although time series of air pressure exhibited a several-day variation due to the passage of synoptic systems, it was not easy to clearly discriminate each migratory anticyclone in Fig. 6(b) due to its weak intensity.The diurnal variation in period mean radiation showed a typical pattern (Fig. 6(c)) that the albedo ranged from 10% to 14% with a mean of 12%.The diurnal variation in the period mean air pressure exhibited a maximum near 0900 LST and a minimum near 1700 LST, with a difference of 2.8 hPa (Fig. 6(d)).This large difference suggests that these variations might be related to local circulations such as the sea-land breeze rather than the atmospheric tide (Frank, 2010).
Generally, surface momentum, sensible heat, and latent heat fluxes are important because they can directly affect the boundary-layer structure and the evolution of the mixinglayer (Stull, 1998;Arya, 1999).Fig. 7 shows the diurnal variation of period mean friction velocity, sensible heat flux, latent heat flux, and CO 2 flux observed at the Jungnang Station during the MAPS-Seoul period.Friction velocity ranged from 0.21 m s -1 to 0.48 m s -1 with a maximum in the late afternoon and a minimum at 0200 LST (Fig. 7(a)).Daily maximum friction velocity in the late afternoon was coincident with the daily maximum wind speed (Fig. 7(c)).Sensible and latent heat flux showed a typical diurnal variation pattern over urban areas.There was a maximum near 1300 LST and small positive values during the nighttime (Park et al., 2014).The positive sensible and latent heat fluxes during the nighttime in an urban residential area were different from the negative values in an urban park or rural area (Park et al., 2013c).The Bowen ratio, the ratio of the sensible heat flux to the latent heat flux, was about 7 (Park and Chae, 2017).The value was larger than that in Melbourne (Coutts et al., 2007), and similar to that in London (Kotthaus and Grimmond, 2014).According to the surface energy balance equation, the higher Bowen ratio meant the larger sensible heat flux.That is, the sensible heat flux in Seoul might be larger than that in Melbourne, and similar to that in London.The ratios of the sensible and latent heat fluxes to the downward shortwave radiation were around 0.28, and 0.04, respectively.Net shortwave radiation (downward solar radiation -upward solar radiation) was found to be around 88% of the downward shortwave radiation.The net radiation (net shortwave radiation + downward longwave radiation -upward longwave radiation) occupied around 70% of the downward shortwave radiation.Consequently, when considering the surface energy balance, the sum of heat storage and anthropogenic heat release occupied as large as 55% of the net radiation.The CO 2 flux exhibited a similar diurnal pattern to anthropogenic emissions, with a maximum value of 1.0 mg m -2 s -1 during the morning rush-hour.The diurnal pattern of CO 2 flux was affected by the anthropogenic energy use such as local heating and road traffic (Kim et al., 2017) as well as the biogenic processes such as photosynthesis, soil and plant respirations besides the meteorological condition such as boundary-layer height.The rush-hour peak was due to the emission from road traffic and less photosynthesis, while afternoon decrease was due to the much photosynthesis by plant in spite of the same road traffic.The CO 2 flux maintained small positive values of between 0.25 and 0.65 mg m -2 s -1 in the remaining hours, which matched the previous results of Park et al. (2014).

Structure of the Atmospheric Boundary-layer
Air pollutants move horizontally with the wind, or are dispersed vertically by turbulent fluxes (Arya, 1999).Atmospheric advection, vertical mixing, and boundarylayer structures are thus key parameters in understanding air pollutant concentration.To this end, boundary-layer structures were investigated using a ceilometer to obtain a vertical profile of attenuated backscatter.Figs. 8(a) and 8(b) show the time-height cross sections of attenuated backscatter during two different periods: one with 6 consecutive clear days, and the other with 4 consecutive cloudy days.In this figure, attenuated backscatters exceeding the value 200 × 10 -8 srad -1 m -1 were substituted to that value in order to exclude the strong attenuated backscatter caused by thick cloud or rain drops.The first event period (23-28 May) consisted of consecutive clear days with two mist/haze events (25 and 26 May).The second event period (9 to 12 June) had 4 consecutive mist/haze event days and rains on 11 and 12 June.The attenuated backscatter depended on the aerosol concentration sensitive to 910 nm wavelength and the vertical aerosol profile represented the boundarylayer structure.So, the boundary-layer features such as mixing-layer height (MLH) and residual-layer could be visually determined from the time-height cross sections of attenuated backscatter.
The MLH was defined as the height up to which atmospheric properties and substances originating from the earth's surface are uniformly dispersed by turbulent vertical mixing process (Emeis et al., 2008).In general, the boundary-layer comprised a single mixing-layer in daytime, but it consisted of two or more layers during the night: a lower mixing-layer and a higher residual-layer as given in Fig. 8(a).Sometimes there was a low attenuated backscatter sub-layer between the two adjacent layers.Attenuated backscatter near the surface is strongly related to the aerosol concentration (Li et al., 2017).Mist/haze events on 25 and 26 May, and 9-12 June were related to the high surface attenuated backscatter which exceeded about 150 × 10 -8 srad -1 m -1 (Figs.8(a) and 8(b)).A series of strong attenuated backscatter signals, below 4 km altitude on 11 and 12 June, represented thick clouds accompanied by precipitation (Fig. 8(b)).
In order to determine the MLH, the following procedures were applied.Firstly, strong attenuated backscatter signals exceeding 500 × 10 -8 srad -1 m -1 were classified as a layer with thick cloud or rain drop, and were excluded (Munkel et al., 2007;Tsaknakis et al., 2011).Secondly, the running means for 11-point temporal (11 minutes) and 11-point vertical (110 meter) data were calculated to minimize the random noise error.Thirdly, the MLH was determined as the height with a minimum vertical gradient of attenuated backscatter (Eresmaa et al., 2006).Finally, ambiguous MHLs, such as those which exhibited two or more maxima of attenuated backscatter within 4 km, were removed.Fig. 8(c) gives an example of the MLH estimated by the vertical profiles of attenuated backscatter during 23-28 May 2015 (Fig. 8(a)).The MLH maintained nearly a constant value in the range of 0.3-0.5 km during the nighttime.It then evolved to maximum value of 1.5-2.5 km during the afternoon (Fig. 8(c)).The ranges, along with its mean and median values, for the nighttime MHL, daily maximum MLH, and daily maximum downward solar radiation are listed in Table 2.The data were excluded from the analysis when the ratio of available data to total data was below 50%.As a result of this stipulation, 19 and 17 days were selected for further analysis on daily maximum and nighttime MLH, respectively.
In order to assess the impact of weather phenomena on MLH, the number of days with each weather phenomenon was investigated.Table 3 shows the number of days where Table 2.The range (with its mean and median) of the nighttime mixing-layer height (0000 LST to 0600 LST), daily maximum mixing-layer height, and daily maximum downward solar radiation obtained at the Jungnang Station during the MAPSSeoul period.

Day (in 2015)
Nighttime mixing-layer height (km) Range (mean, median)  the weather was classified, such as clear and cloudy days, and mist/haze and no-mist/haze event days.It was found that the numbers with each weather classification were relatively evenly distributed in the range between 7 and 12. Fig. 9 shows the case mean and standard deviation of daily maximum downward solar radiation, daily maximum MLH, and mean nighttime MLH.These evaluations were done for the total days, mist/haze days, no-mist/haze days, clear days, and cloudy days, respectively.The daily maximum downward solar radiation had a mean of 800 W m -2 with a standard deviation of 186 W m -2 (Fig. 9(a)).The daily maximum downward solar radiation on the mist/haze event days was 105 W m -2 smaller than on the no-mist/haze event days, while that on the cloudy days was 199 W m -2 smaller than on the clear days (Fig. 9(a)).The standard deviations of daily maximum downward solar radiation on the mist/haze and the cloudy days were larger than those on the nomist/haze and the clear days, respectively (Fig. 9(a)).
Daily maximum MLH height had a mean of 1.77 km with a standard deviation of 0.25 km (Fig. 9(b)).Daily maximum MLH on the mist/haze event days was 160 m lower than that on the no-mist/haze event days.This might be due to the fact that the downward solar radiation on the former was smaller than that on the latter (Fig. 9(a)).Evolution of MLH during the daytime was known to be determined by the accumulated sensible heat flux during the daytime in many meteorological models (Pielke, 2002).So it was common to be positively proportional to the downward solar radiation (Stull, 1988;Park et al., 2014).Daily maximum MLH on the clear days was 140 m higher than on the cloudy days (Fig. 9(b)).This was due to the difference in downward solar radiation, but was not as large as the solar radiation difference of 199 W m -2 .It was found that the mist/haze event had larger effects on the daily maximum MLH than the cloud cover had.
During the MAPS-Seoul period, the nighttime MLH had a mean of 362 m, with a standard deviation of 57 m (Fig. 9(c)).The nighttime MLHs on the no-mist/haze and the clear days were slightly higher than those on the mist/haze and the cloudy days by 10 m and 18 m, respectively (Fig. 9(c)).The mist/haze and cloud cover tended to make the standard deviation of nighttime MLHs smaller than those on the remaining days, even though the standard deviation of downward solar radiation was larger than that on the remaining days (Fig. 9(a)).
Lastly, the MLHs were investigated for the airborne measurement periods.The MLHs exhibited constant values for RF01 (between 1.7 km and 1.8 km), increasing trends for RF03 (1.7-1.9 km) and RF06 (from 1.0 km to 1.3 km), decreasing trends for RF02 (from 1.5 km to 1.2 km) and RF07 (from 1.7 km to 1.0 km).They showed the lowest MLH of 0.7 km for RF04 (morning) and the highest MLH of 1.9 km for RF05.The MLH for RF07 showed the sharp decrease from 1.6 km to 1.2 km around 1530 LST.Airborne measurement exhibited that the organic aerosol concentration abruptly increased at the former height, while the sulfate concentration abruptly decreased at the latter height (NIER, 2015b).

SUMMARY AND DISCUSSIONS
The MAPS-Seoul (Megacity Air Pollution Studies in Seoul) intensive field campaign was carried out during the period from 18 May to 13 June 2015.Its aim was to understand the long-range transport of air pollutants from eastern China, domestic emission processes, and homogeneous or heterogeneous chemical processes in the atmosphere.This study provides an overview of the meteorological surface variables and boundary-layer structures obtained by a surface energy balance system and a ceilometer installed in the SMA.It was found that the MAPS-Seoul period was much warmer and drier, due to a series of migratory anticyclone passage under a strong El Niño event, than the climatological periods from the weather chart and the surface meteorological variables.The wind anomaly analysis showed that the SMA, including the eastern part of the city of Seoul, was affected by the local wind, especially by the sea-land breeze.Because the observed wind included the local wind effects, such as the sea-land, urban-rural, mountain-valley, and river breezes as well as the synoptic system, it was not easy to distinguish the urban-rural circulation from the observed wind.The vertical profile of the attenuated backscatter, obtained by the ceilometer, indicated that the MLH exhibited small values before morning and reached its maximum in the afternoon.The maximum daily MLHs were lower on the misty/hazy and the cloudy days than on the no-mist/haze and the clear days, respectively, due to a decrease in the downward solar radiation.The MLH exhibited a strong relationship with the vertical profiles of sulfate and organic aerosols obtained during the airborne measurement period.
This study suggests that more frequent mist/haze events can induce a lower daytime MLH and increase the air pollutant concentration (Zhao et al., 2016;Yang et al., 2017).However, the analyzed period was too short to generalize and quantify the effects of mist/haze and cloud cover on the MHL.Further research into the long-term effects of mist/haze and cloud cover on the mixing-layer height is needed.More detailed local circulation features, such as the starting and ending time, and the vertical and horizontal regimes of the sea-land, mountain-valley, and urban-rural breezes, should be analyzed in order to gain an improved understanding of the air quality (Bigg et al., 2012;Minoura et al., 2016;Salvador et al., 2016).
The local circulation is driven by thermal differences between the different surface covers, for example, between land and sea, between valley and mountain, and between urban and rural.In addition, the SMA has very complex geography and topography and is composed of many different land covers.This complexity complicates not only the surface-layer turbulence and the boundary-layer structure but also the air quality forecast.
The daytime evolution of the MLH is positively proportional to the surface sensible heat flux and negatively proportional to the lapse rate of the potential temperature above the MLH in the free atmosphere, while the nighttime MLH depends on atmospheric stability and friction velocity.The parameterization of MLH in terms of the sensible heat flux, lapse rate of potential temperature, atmospheric stability, and friction velocity using long-term data requires more sophisticated analysis and is now on hand.

Fig. 1 .
Fig. 1.(a) The locations of the Jungnang and Seoul (KMA) Stations and (b) land cover near the Jungnang Station.The prefixes "R", "M", "C", "G", and "A" indicate residential, mixed, commercial, governmental, and apartment areas, respectively.The radius of 600 m from the Station is shown by a solid circle.

Fig. 2 .
Fig. 2. Time series of hourly mean (a) PM 10 and PM 2.5 , (b) SO 2 and CO, (c) O 3 and NO x concentration, (d) the ratio of PM 2.5 to PM 10 concentration, and (e) visibility in Seoul for the period from 18 May to 13 June 2015.Green columns indicate the airborne measurement periods.

Fig. 4 .
Fig. 4. Time series of (a) air temperature obtained at the Jungnang Station and (b) accumulated precipitation obtained at the Seoul Station for the period from 18 May to 13 June 2015.This is shown with the climatological (a) daily mean, maximum, and minimum air temperatures and (b) accumulated precipitation, as observed at the Seoul Station for 30 years (1981-2010).Green columns indicate the airborne measurement periods.

Fig. 5 .
Fig. 5. Time series of (a) wind speed and (b) wind direction observed at the Jungnang Station for the period from 18 May to 12 June 2015.Diurnal variation of (c) wind speed and direction and (d) their anomalies averaged during the days with a low cloud amount less than 30%.Wind roses during (e) the daytime (1200 LST and 1800 LST) and (f) the nighttime (0000 LST to 0600 LST) for the observation period are also shown.The occurrence frequency of calm wind (wind speed less than 0.5 m s −1 ) is denoted in the center circle in (e) and (f).Green columns indicate the airborne measurement periods.
Figs. 6(a) and 6(b) show a time series of the 4-component net radiation and air pressure obtained at the Jungnang

Fig. 6 .
Fig. 6.Time series of (a) 4-component radiation and (b) air pressure observed at the Jungnang Station for the period from 18 May to 12 June 2015.Diurnal variation of (c) 4 components of radiation and (d) air pressure averaged during the experimental period.Green column indicates the airborne measurement period.

Fig. 8 .
Fig. 8. Time-height cross sections of attenuated backscatter observed by a ceilometer installed at the Jungnang Station for the periods (a) from 23 to 28 May, and (b) from 9 to 12 June 2015, and (c) a time series of the mixing-layer height as estimated by the gradient method using the vertical profile of attenuated backscatter.Research Flight 01 period is indicated in (a) and (c).

Fig. 9 .
Fig. 9. Mean and standard deviation of (a) daily maximum downward solar radiation, (b) daily maximum mixing-layer height, and (c) mean of nighttime mixing-layer height on total days, mist/haze event days, no-mist/haze event days, clear days, and cloudy days.

Table 1 .
Cloud cover (%), weather phenomena at the Seoul Station, and airborne research flight period during the MAPS-Seoul period.Clear days are shaded in light green and mist/haze days are shaded in light blue.The amount of precipitation is inserted in parenthesis.

Table 3 .
The number of weather classification days.